TensorRT quick start, implemented by autodl.
Step1. choose the TensorRT Framework in autodl.
Step2. Install some packages, install what is missing. It depends on your personal needs. You can also refer to my environment in the requirement.txt.
Step1. run the export.py.
Step2. Execute this command on the terminal: trtexec --onnx=model.onnx --saveEngine=model.trt --fp16.
Step3. run the evaluate.py.
run the main.py.
Many thanks to (PriceYH (HaoYu)) for his help!